Index Anomaly Detection Using Gaussian Joint Probabilities
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current index anomaly detection methods in transaction systems, such as third-party payment systems, are not accurate enough due to the influence of jitter from single monitoring points, which can lead to misleading anomaly detection.
Innovation Solution
The method employs a Gaussian model to calculate mean values and variances of monitoring data, then uses these calculations to determine probabilities and joint probabilities across windows of time, allowing for more accurate anomaly detection by considering multiple monitoring points rather than individual points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If anomaly detection is performed on a single monitoring point using historical mean value and variance, then the detection process is simple, but the accuracy is low due to jitter influence
Solution Approach 1:
The patent merges multiple monitoring points into a unified detection framework by calculating joint probabilities across all monitoring points within a time window. Instead of analyzing each monitoring point independently, the system combines their probability distributions and computes the joint probability of anomalies occurring simultaneously, thereby improving detection accuracy while maintaining manageable complexity through mathematical aggregation.
Solution Approach 2:
The patent transitions from single-point anomaly detection to multi-dimensional anomaly detection by introducing the time window dimension. The system evaluates the joint probability distribution across multiple monitoring points within a sliding time window, adding a temporal dimension to the analysis. This dimensional expansion allows the system to distinguish between transient jitter and genuine anomalies more effectively.
2Measurement precision
If multiple monitoring points are considered to improve accuracy, then anomaly detection accuracy improves, but the calculation complexity increases
Solution Approach 1:
The patent changes the mathematical parameters from individual mean and variance calculations to joint probability density function calculations. By transforming the problem into a parameter space where joint probabilities are computed using the multivariate Gaussian distribution, the system achieves more accurate anomaly detection while the computational complexity remains manageable through efficient mathematical formulations.
Solution Approach 2:
The patent performs preliminary calculations of mean values and variances for each monitoring point before computing the joint probability. These preliminary statistics are pre-computed and stored, allowing the main anomaly detection process to directly use these pre-calculated values rather than recomputing them, thereby reducing the overall computational burden while maintaining multi-point analysis capability.
Data Source
AI summary
An index anomaly detection method includes: acquiring data of each of monitoring points, contained in a period of time, of a monitored index; extracting a mean value and a variance of the data of the monitoring points using a Gaussian model; calculating, according to the mean value and the variance of the data of the monitoring points, probabilities of occurrence of the data of the monitoring points, respectively; calculating, according to the respectively calculated probabilities, joint probabilities of occurrence of the data of the monitoring points contained in respective windows divided from the period of time; and detecting, according to the joint probabilities corresponding to the respective windows, whether the monitored index is abnormal.